Shellfish exploitation during the Oakhurst at AUTHORS: Klipdrift Cave, southern Cape, South Africa Kokeli P. Ryano1,4 Karen L. van Niekerk2 Sarah Wurz2,3 Christopher S. Henshilwood2,4 Klipdrift Cave in the southern Cape, South Africa, provides new insights into shellfish harvesting during the Later Stone Age (14–9 ka) period associated with the Oakhurst techno-complex. Two shellfish AFFILIATIONS: dominate: sarmaticus and Dinoplax gigas. An abrupt shift in the relative frequencies of these species 1Department of History, School of Humanities, University of Dodoma, occurs in the middle of the sequence with T. sarmaticus almost completely replacing D. gigas. The shift in Dodoma, Tanzania dominant species is likely due to environmental change caused by fluctuating sea levels rather than change 2SFF Centre for Early Sapiens Behaviour (SapienCE), Department of in sea surface temperatures. The shellfish assemblage shows that local coastal habitats at Klipdrift Cave Archaeology, History, Cultural Studies were somewhat different from those of contemporaneous sites in the southern Cape. Although the shellfish and Religion, University of Bergen, Bergen, Norway specimens are smaller at Klipdrift Cave than those from Middle Stone Age localities such as Blombos Cave, 3School of Geography, Archaeology there is no robust indication that larger human populations at Klipdrift Cave during the Oakhurst period might and Environmental Studies, University have caused this change in size. Environmental or ecological factors could have restricted shellfish growth of the Witwatersrand, Johannesburg, South Africa rates as some experimental works have suggested, but this possibility also remains to be further explored. 4Evolutionary Studies Institute, University of the Witwatersrand, Significance: Johannesburg, South Africa • The dominance of D. gigas and T. sarmaticus at Klipdrift Cave is surprising, as it indicates a local habitat slightly different from other similar sites during the Oakhurst period. CORRESPONDENCE TO: Kokeli Ryano • The shift in dominance from D. gigas to T. sarmaticus indicates changing climatic and environmental conditions during the Oakhurst period, 14–7 ka. EMAIL: [email protected] Introduction DATES: Early evidence for the exploitation of shellfish for subsistence traces back to at least 164 ka during the Middle Stone Received: 11 Sep. 2018 Age (MSA) in South Africa1, and by 100–60 ka shellfish were systematically and intensively exploited at a handful of Revised: 03 Mar. 2019 sites2-5. Evidence for the use of shellfish for purposes other than food, such as making containers and ornaments, Accepted: 16 May 2019 appears from 100 ka to 75 ka in the southern Cape.6-8 It is, however, possible that many older sites containing shellfish Published: 26 Sep. 2019 remains were destroyed by the Marine Isotope Stage 5e sea level transgression.1 Further, there is little evidence for shellfish exploitation between 50 ka and 14 ka, mainly because of the paucity of coastal sites from this time period. HOW TO CITE: Ryano KP, Van Niekerk KL, Wurz Evidence for shell-fishing re-appears at around 14 ka in the southern Cape,9 at the end of the period associated S, Henshilwood CS. Shellfish with the Robberg techno-complex. Shellfish become more abundant in sites during the subsequent period linked exploitation during the Oakhurst at Klipdrift Cave, southern to the Oakhurst techno-complex, around 14–7 ka, although sites from this period with shellfish are still relatively Cape, South Africa. S Afr J Sci. uncommon.10,11 The most abundant evidence for intensive shellfish exploitation comes from the ‘megamidden’ 2019;115(9/10), Art. #5578, period, between 3 ka and 2 ka, from the West coast, which is dotted with extensive open shell middens.12 9 pages. https://doi.org/10.17159/ sajs.2019/5578 Here we present new data on shellfish exploitation during the Oakhurst period from a recently excavated Later Stone Age (LSA) site – Klipdrift Cave (KDC) – situated in the southern Cape. We compare our data to those from ARTICLE INCLUDES: contemporary sites in the region by examining the role that shellfish played in the subsistence of these coastal Peer review ☒ dwellers. Climatic and predation pressure hypotheses are examined against these data. We investigate three main Supplementary material ☒ issues: the nature of marine shellfish exploitation; the intensity of shellfish collection; and the extent to which climate and environmental conditions can be deduced from shellfish remains. DATA AVAILABILITY: Open data set ☐ The Oakhurst period in context All data included ☒ 10,13 On request from author(s) The term Oakhurst techno-complex is used here for sites that typically date to ~14–7 ka and that follow the ☐ Not available Robberg techno-complex, although regional variants occur across southern Africa14. Furthermore, the transition ☐ Not applicable 11 ☐ between entities such as the Robberg and Oakhurst in the Cape region was not synchronous. The Oakhurst is characterised by non-microlithic and bladelet poor lithic assemblages dominated by unstandardised flakes, and EDITOR: frequent use of coarser grained lithic raw materials.15-17 Formal tools are rare and consist mainly of medium to Maryna Steyn large scrapers. Non-lithic artefacts associated with the Oakhurst include worked bone, ostrich eggshell beads and ornaments as well as worked marine shell and beads.10 The Robberg period is characterised by relatively KEYWORDS: few shellfish remains, but during the Oakhurst, shellfish, fish, marine mammals and seabirds are present in the shell-fishing, predation pressure, palaeoclimate, environment, Terminal deposits. Initially the Oakhurst people also hunted large-to-medium-sized game such as eland and warthog, but 18 Pleistocene, coastal subsistence these prey were later replaced by browsers, likely reflecting a change towards a woodier habitat.

FUNDING: Shellfish remains during the Oakhurst Complex on the southern Cape coast Research Council of Norway, Of the eight securely dated Oakhurst sites (Figure 1a) in the southern Cape10, four – Nelson Bay Cave (NBC)9, Matjes University of Bergen, University River Shelter (MRS)19, Byneskranskop 1 (BNK 1)20, and Oakhurst Shelter21 – contain significant shellfish remains, of the Witwatersrand, National 15 Research Foundation (South Africa), probably because they were close to the coast. At inland sites such as Boomplaas Cave (BPA) and Kangkara Cave, Palaeontological Scientific Trust Wilton Large Rock Shelter22 and Melkhoutboom Cave23, shellfish are rare, and, when present, may have predominantly (PAST), European Research Council been used for non-subsistence purposes such as the manufacture of ornamental and decorative items. The shellfish from three of the four sites have been described to varying degrees. The shellfish from Oakhurst Shelter were listed to species level but not quantified. The impression is that the shellfish samples were not retained during © 2019. The Author(s). Published excavation.24 It is, however, evident that the white sand mussel (Donax serra) is the most common species in the under a Creative Commons Oakhurst levels.21 Therefore, NBC, MRS and BNK 1 are described in more detail below. Attribution Licence.

Research Article Volume 115| Number 9/10 https://doi.org/10.17159/sajs.2019/5578 1 September/October 2019 Oakhurst at Klipdrift Cave and shellfish exploitation Page 2 of 9

18°0'0"E 19°0'0"E 20°0'0"E 21°0'0"E 22°0'0"E 23°0'0"E 24°0'0"E 25°0'0"E a

33°0'0"E

34°0'0"E

35°0'0"E

M.A.S.L b 35

30

25

20

15

10

5

SURFACE JY cal BP 10760-11170 JYA JZ JZA cal BP 12735-13000 JZB KAB cal BP 13405-13560 KAC KAD cal BP 13375-13555

cal BP 13480-13740 c KAE

Figure 1: (a) Sites containing material from the Oakhurst Complex in the southwestern Cape, (b) Klipdrift Complex (Klipdrift Cave is the western section and Klipdrift Shelter is to the east), and (c) part of the Klipdrift Cave natural stratigraphic profile and associated accelerator mass spectrometry dates (dates provided by Beta Analytics).

NBC and MRS are situated approximately 10 km apart with similar shellfish table, they were not convinced that they were brought in rocky shores and sandy beaches in close vicinity. The most common primarily as food, but suggest that they had been used as artefacts, as species in the Oakhurst levels are the bivalves Perna perna (brown some perforated fragments were found.20 Modified D. serra shells have mussel), D. serra and Choromytilus meridionalis (black mussel). At both not been reported from NBC and MRS. sites, there is an inverse correlation in the frequency of these species over time.9,19 At NBC, C. meridionalis is replaced by P. perna around Shellfish as a proxy for human population sizes 9,11 10 ka whilst D. serra frequencies increase at the same time. At MRS, It has been argued that the sizes of gastropods, and variation in shellfish 13 P. perna increases sharply relative to C. meridionalis by 9.5 ka , but, species, from archaeological sites may provide an estimate of the extent unlike at NBC, D. serra is most common in the earlier layers, dated to and intensity of harvesting of these as food, which in turn can be 19 10 ka, and decreases as P. perna increases . These changes in species used as a reflection of human population size.4,25,26 The reduced size of representation over time have been interpreted as indicative of changing gastropods in LSA sites relative to those from MSA localities in southern 9,19 environmental conditions. Africa has been used to imply intensified collection due to higher human At BNK, shellfish are present in the Oakhurst layers in relatively low population size during the LSA.4,25,27 In addition, MSA assemblages tend frequencies; the total minimum number of individuals (MNI) is 310. to contain a smaller range of mostly larger species, and the few smaller The frequencies gradually increase from the oldest 13 ka to the youngest species tend to occur in low numbers relative to LSA assemblages.26 7 ka Oakhurst layers, but only become significantly abundant in the As such, gastropod size and species abundance have been used to overlying layers attributed to the Wilton period.11,20 The low numbers argue for smaller human population sizes during the MSA.4,25,27 Another probably reflect an increased distance from the shore prior to the Wilton. argument is that the reduced gastropod size could be due to environmental Numbers of C. meridionalis and P. perna are negligible, although the factors affecting shell growth rates, particularly as non-food species are former outnumbers the latter in all the Oakhurst layers. D. serra is the also reduced in size in the LSA compared to MSA.28 It is also interesting most common species in all but the youngest Oakhurst layer, dated that data on Nerita tessellata (tessellated nerite snail) from the Caribbean to 7 ka, where Turbo sarmaticus (alikreukel) become more common. island of Nevis indicate that shellfish size increased despite intensive Although the authors listed non-modified D. serra under the general exploitation by human populations between 890 and 1440 AD.29

Research Article Volume 115| Number 9/10 https://doi.org/10.17159/sajs.2019/5578 2 September/October 2019 Oakhurst at Klipdrift Cave and shellfish exploitation Page 3 of 9

Site background (periwinkles) are present, but, as the apices are usually separate from the identifiable body , the shell weights and MNI have been combined KDC (34o27.0963’S 20o43.4582’E) is located on the coast in the for these two species and listed as Diloma spp. No shell fragments of De Hoop Nature Reserve on Portion 20 of farm 516, Swellendam District D. variegata were found, and it is therefore assumed that only the former in the southern Cape, South Africa (Figure 1a and 1b). It forms part of two species are represented by the apices of this family at KDC. All the the Klipdrift Complex where Klipdrift Shelter, a Howiesons Poort locality, species identified occur in the southern Cape today,33 and no cold-water is also found.5 KDC was first excavated in 2010 and subsequently indicator species (e.g. Cymbula granatina, granite limpet) are present. in 2011. The surface of the deposit in KDC is truncated possibly by mid-Holocene higher sea levels, but excavations over an area of All species listed in Table 1 are edible, and most were presumably collected 2.75 m2 revealed horizontal in-situ depositional layers with exceptional primarily as food. It is possible that the white mussels (D. serrra) were first preservation of bone, shell, charcoal and ostrich eggshell. The lithics are eaten, and some shells subsequently used for other purposes, as 14 of the assigned to the Oakhurst techno-complex, and the top and bottom of the valves have ~10 mm circular perforations near the centre. The angular excavated sequence have been dated to approximately 10 ka and 13 ka, surf clam, Scissodesma spengleri, present in small numbers throughout respectively (Figure 1c).13 the sequence, may also have been used for purposes other than food.34 This species occurs subtidally in the deeper surf zone, and is therefore Materials and methods difficult to collect live but specimens do wash up after storms.35 The KDC All the shellfish remains retained in the 3-mm sieve from layers JY through specimens do not appear waterworn, but, according to G. Branch (2013, KAE, which form 85% of the total excavated volume (0.93 m3) at KDC, were written communication, May 15), these washed up shells are seldom analysed. Whole shellfish <2 cm are not considered to be food items30 but damaged or waterworn. Thus, it is not clear whether these specimens rather animals that landed up in the site incidentally, for example through were collected dead or alive. Some of the valves have what appears to attachment to bigger shells, and are not included in this analysis. be retouch on the ventral side, and might have functioned as a sort of scraper, but this possibility needs further investigation. Incidental, non- The methods and techniques adopted for analysing marine shellfish food species consist mostly of barnacle fragments and juvenile limpets. from KDC involve species identification, determining the MNI, weighing the shells, and measurements of the maximum ‘length’ of T. sarmaticus Shellfish exploitation through time opercula and limpet shells. Both MNI and weight are used as rare species may be underrepresented when only MNI is used31 and further because T. sarmaticus and D. gigas are the most frequently occurring species post-depositional damage can affect the MNI counts32. throughout the assemblage (Tables 1 and 2), contributing over 93% in terms of MNI and 95% in terms of weight. All other species combined MNI counts for T. sarmaticus are derived from counting apices and opercula contribute <4% in terms of weight, and 7% in terms of MNI to the and the highest value is considered the MNI. The weight for T. sarmaticus total assemblage (Table 2). P. perna, midae (), Haliotis species given here includes the opercula and shell weights. The MNI values spadicea (Venus ear, a small abalone), Scutellastra longicosta (long- for other gastropods are calculated by counting the apices. For Dinoplax spined limpet) and S. spengleri (surf clam) occur in negligible numbers. gigas (the giant chiton), the front, middle (the number of middle valves Cymbula oculus (goat’s eye limpet), Diloma spp. and Burnupena cincta divided by six), and rear valves were counted separately. The greatest total (whelk) occur in slightly higher numbers than the aforementioned, but for the three categories was taken as the MNI. Left and right hinges of still at very low frequencies relative to D. gigas and T. sarmaticus. bivalves were counted separately, and the highest value taken as the MNI. D. gigas is the most abundant species in the site, both in terms of weight Results and MNI (Table 1). There is an inverse relationship in frequency between D. gigas and T. sarmaticus through time, with the former being most Species composition abundant in the lower part of the sequence (layers KAE–JZB), and the Eleven mollusc species with a total MNI of 5330 were identified from latter in the upper four layers (JZA–JY). On a much smaller scale, the 197.69 kg of shellfish remains (Table 1). Both Diloma sinensis and D. tigirina frequencies of B. cincta, C. oculus and Diloma spp. follow a similar

Table 1: Klipdrift Cave: Minimum number of individuals (MNI) and weight (g) of shellfish from the Oakhurst layers

Layer Species JY JYA JZ JZA JZB KAB KAC KAD KAE TOTAL MNI g MNI g MNI g MNI g MNI g MNI g MNI g MNI g MNI g MNI g Turbo 279 16 477 167 4959 280 19 605 196 15 889 14 534 22 735 17 520 44 2 168 32 2280 1 051 63 167 sarmaticus Dinoplax 53 906 83 1598 137 2843 85 1896 25 397 107 1999 132 3261 1771 75 142 1527 39 509 3920 127 551 gigas Diloma spp. 3 145 10 405 63 1330 13 453 1 7 1 8 1 37 3 29 1 8 96 2422 Cymbula 33 449 31 691 46 901 2 69 1 2 1 7 1 4 1 4 1 4 117 2131 oculus Scutellastra 2 11 1 7 1 20 4 38 longicosta Perna perna 4 10 1 1 1 1 1 1 1 3 1 4 9 20 Haliotis 1 3 1 2 1 5 1 6 4 16 midae Haliotis 1 2 1 2 spadicea Burnupena 1 1 1 1 1 7 1 16 6 72 14 171 24 328 17 428 65 1024 cincta Donax serra 7 163 4 30 9 128 8 76 1 27 10 158 3 103 5 130 2 117 49 932 Scissodesma 1 5 1 1 1 5 1 6 1 4 1 31 1 55 6 208 1 73 14 388 spengleri Total 382 18 159 300 7698 539 24 821 309 18 423 45 988 150 3019 170 4155 1854 78 009 1581 42 419 5330 197 691

Research Article Volume 115| Number 9/10 https://doi.org/10.17159/sajs.2019/5578 3 September/October 2019 Oakhurst at Klipdrift Cave and shellfish exploitation Page 4 of 9

Table 2: Klipdrift Cave: Relative abundance (%) of each species per layer based on minimum number of individuals (MNI) and weight (g)

Layer Total Species JY JYA JZ JZA JZB KAB KAC KAD KAE MNI g MNI g MNI g MNI g MNI g MNI g MNI g MNI g MNI g MNI g Turbo sarmaticus 73.0 90.7 55.7 64.4 51.9 79.0 63.4 86.2 31.1 54.0 14.7 24.3 10.0 12.5 2.4 2.8 2.0 5.4 19.7 32.0 Dinoplax gigas 13.9 5.0 27.7 20.8 25.4 11.5 27.5 10.3 55.6 40.2 71.3 66.2 77.6 78.5 95.5 96.3 96.6 93.1 73.5 64.5 Diloma spp. 0.8 0.8 3.3 5.3 11.7 5.4 4.2 2.5 2.2 0.7 0.7 0.3 0.6 0.9 0.2 0.0 0.1 0.0 1.8 1.2 Cymbula oculus 8.6 2.5 10.3 9.0 8.5 3.6 0.6 0.4 2.2 0.2 0.7 0.2 0.6 0.1 0.1 0.0 0.1 0.0 2.2 1.1 Scutellastra longicosta 0.0 0.0 0.7 0.1 0.2 0.0 0.3 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 Perna perna 1.0 0.1 0.3 0.0 0.2 0.0 0.0 0.0 2.2 0.1 0.7 0.1 0.6 0.1 0.0 0.0 0.0 0.0 0.2 0.0 0.3 0.0 0.3 0.0 0.0 0.0 0.3 0.0 0.0 0.0 0.7 0.2 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 Haliotis spadicea 0.0 0.0 0.0 0.0 0.0 0.0 0.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Burnupena cincta 0.3 0.0 0.0 0.0 0.2 0.0 0.3 0.0 2.2 1.6 4.0 2.4 8.2 4.1 1.3 0.4 1.1 1.0 1.2 0.5 Donax serra 1.8 0.9 1.3 0.4 1.7 0.5 2.6 0.4 2.2 2.7 6.7 5.2 1.8 2.5 0.3 0.2 0.1 0.3 0.9 0.5 Scissodesma 0.3 0.0 0.3 0.0 0.2 0.0 0.3 0.0 2.2 0.4 0.7 1.0 0.6 1.3 0.3 0.3 0.1 0.2 0.3 0.2 spengleri Total 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 pattern: B. cincta is most common in the lower layers associated with D. gigas and all but disappears in the upper layers, whereas the relative frequencies of C. oculus and Diloma spp. increase in the upper layers (Figure 2). D. serra is present in all layers in low numbers, but its relative frequency is highest in the same layers where B. cincta is most common.

100 90 80 70 60 B. cincta % 50 C. aculus 40 Figure 3: Density of shellfish per layer (kg/m3) in Klipdrift Cave. 30 Diloma spp. 20 T. sarmaticus 10 Shellfish size D. gigas 0 JY JYA JZ JZA JZB KAB KAC KAD KAE Turbo sarmaticus opercula The southern Cape species most frequently used for size measurements Figure 2: Relative abundance (%) per layer of the five most common are various limpets and the opercula of T. sarmaticus. The latter are used shellfish species, based on weight, from Klipdrift Cave. as a proxy for shell size as shells tend to be fragmented in archaeological assemblages.37 Descriptive statistics for T. sarmaticus opercula from P. perna are absent from the lowermost two layers and layer JZA, and KDC are given in Table 3 and summarised using box plots (Figure 4). constitute only between 0.2% and 2.2% of MNI in the layers in which they do occur (Table 2). H. midae and H. spadicea occur in negligible Table 3: Klipdrift Cave: Turbo sarmaticus opercula descriptive statistics quantities, but it is notable that they are only present in layers above JZB, per layer (mm) except for a few fragments in KAB (Table 2). KDC contains a limited number of species (n=11) relative to the other sites, Layer n Minimum Maximum Mean Median s.d. particularly MRS (n=20). Some shellfish species, such as the Aulacomya JY 94 18 46 32.4 33 6.23 atra (ribbed mussel), C. compressa (kelp limpet) and C. granatina, are JYA 50 10 46 31.7 32 6.69 restricted to one site (MRS). Limpets are rare at KDC and BNK 1 and more common at NBC and MRS. S. spengleri is present only at KDC and BNK 1 JZ 114 20 44 33.8 34 5.61 (Supplementary table 1). JZA 52 17 48 35.4 36.5 6.43 JZB 7 26 41 34.7 36 5.19 Shellfish density through time KAB 7 13 36 28.3 30 7.23 Shellfish densities at KDC are very high in the three uppermost and KAC 5 14 38 29.8 33 9.83 two lowermost layers of the sequence (Figure 3). Densities are the KAD 13 24 47 37.1 39 6.55 lowest between layers KAC and JZA. Layer KAD has the highest shell KAE 6 16 45 32.3 35 11.13 density (~374 kg/m3) and JZB the lowest, at ~28 kg/m3. Jerardino36 cautions against using density measures particularly when making inter- site subsistence comparisons. However, as KDC is a ‘closed’ cave (as Opercula lengths range between 10 mm and 48 mm through the sequence. opposed to open air sites), intra-site density comparisons are less likely The median value is highest in layer KAD, at 39 mm, and lowest in KAB to be significantly problematic, although deposition rates may have differed (30 mm) (Table 3). The data for opercula are also presented using box between layers. plots (Figure 4). This figure shows some variations within the sequence.

Research Article Volume 115| Number 9/10 https://doi.org/10.17159/sajs.2019/5578 4 September/October 2019 Oakhurst at Klipdrift Cave and shellfish exploitation Page 5 of 9

For example, larger individuals are apparent in layers KAD and JZA while The (non-food) species such as Cellana radiata capensis and Alaba KAB and JYA have proportionally smaller individuals. pinnae indicate warmer waters38 but these species do not occur at KDC. The species conventionally used as temperature indicators – C. meridionalis for ‘mostly cool’ temperatures and P. perna and C. oculus for ‘mostly warm’ sea surface temperatures – cannot be regarded as reliable proxies for sea surface temperatures.38 An experimental study has suggested that C. meridionalis do not thrive in temperatures above 18 °C39, but C. meridionalis can co-exist with P. perna in the south coast surviving in sea surface temperatures above 20 °C40. This supports the Langejans et al.38 suggestion that C. meridionalis and P. perna may not be reliable temperature indicators. Furthermore, there are minor differences in habitat preferences between C. meridionalis and P. perna that may cause them to co-exist spatially separated in the same locality. C. meridionalis, for example, occurs on rocks on the low shore that are associated with sand while P. perna occurs on the high shore on rocks which are not usually covered by sand.40

Figure 4: Box plots summarising the lengths of opercula of Turbo As most of the species that occur at KDC thrive in both warm and cool sarmaticus from Klipdrift Cave through time. Centre lines show sea temperatures (Table 4), it is difficult to infer sea surface temperatures the medians; box limits indicate the 25th and 75th percentiles at the times of occupation. However, the absence of C. granatina, a more 35,38 as determined by R software; whiskers extend 1.5 times the reliable cold-water indicator species at KDC, NBC and MRS suggests interquartile range from the 25th and 75th percentiles; outliers that sea surface temperatures in the southern Cape coast were mildly are represented by dots. Number of sample points per layer is warm during the Oakhurst period. given on the left. (Constructed using BoxPlotR at http://shiny. chemgrid.org/boxplotr/). Table 4: Shellfish species present at Klipdrift Cave and sea surface temperature38 KDC measurements were compared with those from other LSA localities (Blombosfontein [BBF], Blombos Cave [BBC], and NBC) and MSA sites Species Sea surface temperature (Klasies River [KR], Klipdrift Shelter [KDS], and BBC) from the southern Dinoplax gigas Warm and cool Cape (Supplementary table 2; Supplementary figure 1). BBF 2 specimens are slightly larger than those from both KDC and NBC sites while there is Turbo sarmaticus Warm and cool a progressive decrease in size of Turbo opercula for younger LSA sites. Diloma sinensis Warm and cool The difference in size of specimens between Oakhurst and post-Oakhurst Diloma tigrina Warm and cool assemblages such as BBF 3 is significant (Supplementary figure 1). Cymbula oculus Mostly warm The general trend is that specimens from MSA sites are larger than those from LSA ones; for example, there is a significant difference between BBC Perna perna Mostly warm M2 and KDC (Supplementary figure 1). Haliotis midae Warm and cool Haliotis spadicea Warm and cool Cymbula oculus shells Burnupena cincta Warm and cool The most common limpet species present at KDC is C. oculus. As whole (measurable) C. oculus shells were rare (n=17), the measurements Donax serra Warm and cool were combined for all layers (Supplementary table 3). While the sample Scissodesma spengleri Mostly warm size is small, we include the data here as a contribution to the available information on shellfish size patterns during the Oakhurst period in the southern Cape. As with the T. sarmaticus measurements, C. oculus Species representation can also reflect past habitats. In this regard it measurements from KDC were plotted against those from other MSA is surprising that C. meridionalis is not present, even in the lower KDC and LSA sites in the southern Cape (Supplementary figure 2). C. oculus sequence, as the dominant presence of D. gigas suggests sand inundated specimens from KDC are smaller in size than those from NBC Oakhurst rocky shores – a habitat that is attractive to C. meridionalis. This species layers (Supplementary table 3) using the median index (66.5 mm against is also present at the other Oakhurst sites mentioned. The low incidence of 62 mm). C. oculus specimens from MSA sites such as KR and BBC are sessile mussels such as P. perna at KDC (Tables 1 and 2) is also unusual larger than those from KDC by at least 7 mm (Supplementary figure 2). as they are typically common in LSA sites of the southern Cape coast such as NBC9, MRS19 and the BBF sites (dating from ca 6 ka to 0.5 ka)32. Discussion The fluctuating presence of D. gigas and T. sarmaticus at KDC may be due We address two main issues here: first, whether the change in shellfish to changes in the habitat best suited to each species over time. The shift species over time at KDC could be attributed to climate, environment and/ from the dominance of the more sand-tolerant species, D. gigas, in the or human choices; and second, whether human agents were responsible lower layers (12 ka) to the dominance of T. sarmaticus in the upper layers for the differences in the shellfish sizes observed at KDC. may suggest scouring out of sand in the later period. T. sarmaticus would have thrived in a habitat with more exposed rocks and less sand. The near Change of shellfish species composition at Klipdrift Cave absence of sessile mussels at KDC may suggest a sheltered sandy bay The KDC data present two clear patterns of exploitation: (1) the in front of the cave at times – an environment not favoured by these dominance of D. gigas in the lower layers, with AMS dates centring on species.41 The slight increase in sessile mussels towards the top of the 14 ka and 13 ka and (2) the high frequency of T. sarmaticus in the upper sequence could indicate a change to rockier shores and rock pools that layers, from layer JZA (Figures 2 and 3). The question we explore here would also have attracted T. sarmaticus and limpets such as C. oculus.41 is whether this shift in the presence of species is related to changes in The subtle changes in coastal morphology suggested by the shifting sea surface temperatures, habitat change or deliberate human choice. dominance of the species may have been a result of rising sea levels Shellfish species composition has often been used as an indicator of and is less likely due to changing sea surface temperatures. The rising sea surface temperatures.2,9,19 However, only a few species are effective sea levels signalled the transition from the Last Glaciation towards temperature indicators. These species include C. granatina and A. atra35,38 the Holocene epoch. The complete absence of H. midae and the low that occur on the west coast and are indicative of cool temperatures. incidence of P. perna and C. oculus (species that do not tolerate overly

Research Article Volume 115| Number 9/10 https://doi.org/10.17159/sajs.2019/5578 5 September/October 2019 Oakhurst at Klipdrift Cave and shellfish exploitation Page 6 of 9

sandy environments) in the ca 10 ka layers42 support a scenario of a sandy and the turban shells in the MSA and LSA sites, may have been caused dominated marine environment around this time. by a combination of both natural and human factors.28 A final scenario to consider for the change of species composition at Non-human factors that affect the shellfish growth rates include sea KDC is whether this change relates to human choice, acknowledging that surface temperature and turbidity, salinity, topography, wave action, it is complicated to discriminate between changes resulting from human desiccation, shellfish population densities and food supply.28,44,47-49 choice and those from the environment.28,43 Although the dominant Oceanic productivity or the production of organic matter by phytoplankton, species at KDC prefer slightly different habitats, it is common for them generates food for marine life such as shellfish.50 Productivity changed to, at times, occur in close vicinity, suggesting that both could have been over time and it is known that the primary productivity of the Subantarctic available for collection during gathering events. Ocean changed over the last 70 ka with marked algal production at ca 58.8, 53, 46 and 38.5 ka.51 Oceanic productivity data for the southern One possible indication that human choice was responsible for the Cape coast are not available, but productivity may have been influenced difference in representation through time is the size of T. sarmaticus in the by the Subantarctic Ocean.28 Variations in oceanic primary productivity lower layers. If the coastal zones were newly colonised by this species affect the food chain and, in turn, may affect size and distributions of in the lower layers, then one would expect the population to consist of shellfish species.28 The growth of T. sarmaticus is affected not only by smaller animals, migrating from crowded subtidal populations, not the lack of food but also by its quality.44 Changes in oceanic productivity larger ones that tend to stay in the lower subtidal areas.33,44 While Turbo may have resulted in changes in the availability and the quality of food opercula measurements in the lower layers at KDC (Table 3) indicate a on the southern Cape coast, although this supposition remains to be relatively small average size, there are some large individuals present, particularly in layers KAE and KAD (Figure 4). The presence of such firmly established. large individuals suggests that a mature Turbo population was present In the case of the KDC data, to test whether increased predation led to a and available and could be tentative evidence that people actively chose decrease in size, we predicted size reduction in T. sarmaticus from the to collect D. gigas in the older levels, despite the availability of good- older to the younger layers, when exploitation of this species intensified. sized T. sarmaticus. However, it seems more convincing at present This prediction is based on the premise that a present but unexploited to suggest that the shifting dominance of D. gigas and T. sarmaticus Turbo community will contain many large specimens.52,53 It has also at KDC was caused by habitat change rather than human preference. been hypothesised that humans tend to target the largest specimens A similar scenario is suggested for the MSA site of KDS at the same first when gathering shellfish and the smaller ones may be collected later locality, where D. gigas replaces T. sarmaticus and H. midae in the upper and thus the overall size distribution would become skewed.3,4,25,27,54-56 layers.5 This argument may be tested by future research when refined If shellfish collectors were intentionally seeking out a species at KDC, one palaeoenvironmental reconstructions of KDC become available. In the would expect the initial assemblage to contain the largest specimens, instance of C. meridionalis and P. perna, it is unlikely that people would and a gradual reduction in size through time as increased predation leads discriminate between the former and the latter when collecting, as they to fewer large specimens being available.3,4,27,56 are presumably the same in terms of size and taste.9 Thus, the absence of C. meridionalis and the rarity of P. perna may be most likely explained Opercula measurements show that sizes decrease from layer JZA by environmental factors rather than human decision to not collect them. upwards, and the difference in size between JZA and the uppermost two layers, JYA and JY, is statistically significant.13 Thus, the decrease in size Cause(s) of size reduction through time of T. sarmaticus opercula at KDC, especially after JZA, may Shellfish sizes in archaeological sites have been linked to human support a scenario of intensive exploitation leading to reduction in size. population sizes and the intensity of harvesting. Here, the comparison This decrease coincides with an increase in shellfish densities, which of shellfish size between MSA and LSA sites in the southern Cape could be because of more intensive harvesting or occupation intensity is discussed only for T. sarmaticus opercula and C. oculus shell at this time. However, this does not explain why the KDC opercula are measurements. Given the relative rarity of C. oculus at KDC, their overall smaller than those in MSA contexts. It is possible that T. sarmaticus at KDC small size (Supplementary table 3) is unlikely caused by human predation had slower growth rates than during the MSA due to not yet established pressure. There is no criterion established for comparing D. gigas sizes environmental factors. Although T. sarmaticus were rare in the lower part although they are numerous at KDC45, KDS and BBC MSA38,46 and are of the sequence at KDC, when presumably little exploitation occurred, also present at MRS, between 9.6 ka and 7 ka19. Comparing their size they are still smaller than MSA ones, which suggests that environmental through time may be a subject for later research. conditions affected their growth rates. As detailed above, KDC T. sarmaticus opercula are smaller in size than Although there were probably larger human populations during the those from the MSA of KDS, BBC and KR sites, but larger than most post- LSA25, non-human factors could also have impacted shellfish size43. Oakhurst assemblages from BBC, KR and BBF (Supplementary figure 1). Reduced size of shellfish may also be a function of more frequent The few measurable C. oculus at KDC are also smaller than those from harvesting by smaller groups.43 Until all the causal factors are carefully MSA sites and more like those from the Oakhurst layers at NBC. weaved together, larger human population as the only driver of shellfish size reduction is untimely.43 Although most analysts of the southern Cape coast molluscs3,4,25,27 argue that intensive harvesting of shellfish due to larger human population is a The environment and the Oakhurst subsistence economy leading cause of reduction in average size of marine molluscs through Shellfish remains are rare during the Robberg, a period that precedes the time, others28,43,47,48 question this proposition. Jerardino et al.43 and Sealy Oakhurst (e.g. at NBC57), as sea levels were lower during the Last Glacial and Galimberti28, for example, point out that C. meridionalis in MSA and Maximum. Shellfish become abundant again from the Oakhurst period and LSA occurrences are similar in size while limpets (e.g. C. oculus and thereafter. The increase in shellfish subsistence during the Oakhurst period S. argenvillei) and turban shells (e.g. T. sarmaticus) are smaller in the LSA. coincides with the rise of sea levels. The sea level transgression after Klein and colleagues4,25,27 believe that the lack of significant difference in 14.5 ka58 brought the coastline very close to the present-day Oakhurst C. meridionalis sizes between MSA and LSA sites on the west coast is sites on the southern Cape coast. The shellfish species exploited at KDC due to this species’ rapid colonisation relative to that of slower growing differ somewhat from those at MRS and NBC, and changes through time gastropods or turban shells such as T. sarmaticus. are evident at all three sites. At KDC, for example (Figure 2), there is a It may be significant that non-food shellfish such as Nassarius change from the dominance of D. gigas to T. sarmaticus in the sequence kraussianus (tick shell) from the MSA at BBC are significantly larger after/around 12 ka (from layer JZA), the period that coincides with the than those from the LSA levels at the same site and at Die Kelders.28 driest environment in the sequence as suggested by ostrich eggshell It is unlikely that the reduced sizes of N. kraussianus in LSA contexts isotopes.45 At MRS and NBC, P. perna replaces C. meridionalis at about can be attributed to intensive collection as they were not that intensively 10 ka. These changes are likely a result of changes in local habitat collected.28 Hence, the differences in sizes of shellfish, especially limpets conditions through time and site-specific shores. The isotopic data from

Research Article Volume 115| Number 9/10 https://doi.org/10.17159/sajs.2019/5578 6 September/October 2019 Oakhurst at Klipdrift Cave and shellfish exploitation Page 7 of 9 ostrich eggshells also indicate maximum aridity in the sequence in layer Acknowledgements JZA.45 Furthermore, shellfish density is lowest in layer JZB (Figure 3), Partial funding for this research was provided to C.S.H., K.L.v.N. and S.W. which also shows a decrease in lithic production. When these trends are by the Research Council of Norway through its Centres of Excellence compared to the temperature data for the terrestrial sequence of pollen at Wonderkrater59 (Thackeray and Scott’s59 Figure 2 reproduced here as funding scheme, Centre for Early Sapiens Behaviour (SapienCE) (project Figure 5), it is clear that layer JZB, where the lowest density of shellfish is number 262618). C.S.H. was also funded through a South African recorded, coincides with the time when temperatures were probably the National Research Foundation Research Chair (SARChI) at the University lowest during the Younger Dryas. The Younger Dryas event might indeed of the Witwatersrand and by the University of Bergen, Norway. S.W.’s have had a cooling effect over environments in southern Africa, but the research was further supported by the National Research Foundation of influence of this effect may have varied regionally as highlighted by Fitchett South Africa (grant no. 98826). Additional funding was provided to C.S.H. and colleagues60. On the other hand, the lowest densities of both shellfish by the European Research Council (ERC) under the European Union’s and lithic artefacts at layer JZB may also suggest a low-occupation period Seventh Framework Programme (FP7/2007-2013)/ERC grant agreement at the cave, but this argument needs to be supported by other faunal data. no. 249587 and by the Palaeontological Scientific Trust (PAST), Johannesburg, South Africa. Opinions expressed and conclusions arrived at are those of the authors and not those of the funding agencies. Petro Keene and Samantha Mienies assisted during analysis while Richard Klein generously supplied data and created Supplementary figures 1 and 2 of this work. Magnus Haaland and Ole Frederik Unhammer kindly supplied some of the images in Figure 1. Authors’ contributions K.P.R. contributed to data analysis, the methodology and writing of the initial and revised drafts. K.L.v.N. contributed to data collection and analysis, the methodology and writing of the revised draft and co-directed excavations at the Klipdrift Cave site. S.W. was involved in data collection, data analysis and methodology, student supervision and writing revisions. C.S.H. conceptualised the research project, acquired funding, and provided project leadership and management and student supervision. Figure 5: Temperature index for Wonderkrater within the past 16 000 years (calibrated dates) showing layers KAE, JZB and JY at References KDC. Layer JZB, for which the shellfish density is the lowest, 1. Marean CW, Bar-Matthews M, Bernatchez J, Fisher E, Goldberg P, Herries AIR, coincides with the time when temperatures might have been et al. Early human use of marine resources and pigment in South Africa during relatively low within the Younger Dryas. Figure modified after the Middle Pleistocene. Nature. 2007;449:905–908. https://doi.org/10.1038/ nature06204 Thackeray and Scott’s59 Figure 2, with permission. 2. Thackeray JF. Molluscan fauna from Klasies River, South Africa. S Afr Archaeol The change in shellfish composition probably reflects habitat change Bull. 1988;43(147):27–32. https://doi.org/10.2307/3887610 that involved removal of sand from the rocks after or around 11 ka due 3. Klein RG, Avery G, Cruz-Uribe K, Halkett D, Parkington JE, Steele T, et al. to increasing sea levels. The decrease of the sand mussels at MRS and The Ysterfontein 1 Middle Stone Age site, South Africa, and early human NBC in the upper layers also supports the shrinking of sandy shores at exploitation of coastal resources. Proc Natl Acad Sci USA. 2004;101(16):5708– this time. It is noteworthy that the shellfish subsistence practices track 5715. https://doi.org/10.1073/pnas.0400528101 the change in environment much more closely than the lithic technology does at KDC.13 The lithic technology remains stable. 4. Avery G, Halkett D, Orton J, Steele T, Tusenius M, Klein RG. The Ysterfontein 1 Middle Stone Age rock shelter and the evolution of coastal foraging. Goodwin Conclusions Ser. 2008;10:66–89. As there are only a few sites in the southern Cape with exceptionally 5. Henshilwood CS, Van Niekerk KL, Wurz S, Delagnes A, Armitage SJ, Rifkin preserved shellfish remains, this paper broadens our understanding and RF, et al. Klipdrift Shelter, southern Cape, South Africa: Preliminary report on provides new data on shell-fishing during the Oakhurst period. There are the Howiesons Poort layers. J Archaeol Sci. 2014;45:284–303. https://doi. org/10.1016/j.jas.2014.01.033 inter- and intra-site variations in the shellfish species exploited – perhaps because of subtle habitat changes. 6. Henshilwood CS, d’Errico F, Vanhaeren M, Van Niekerk K, Jacobs Z. Middle Stone Age shell beads from South Africa. Science. 2004;304:404. https://doi. Of the 11 mollusc species that occur at KDC, 2 are dominant: D. gigas org/10.1126/science.1095905 is abundant in the lower layers while T. sarmaticus is more numerous upwards. The density of shellfish at KDC is lower in the middle of 7. d’Errico F, Vanhaeren M, Wadley L. Possible shell beads from the Middle Stone the sequence, which may be due to sea level regression and/or Age layers of Sibudu Cave, South Africa. J Archaeol Sci. 2008;35:2675–2685. https://doi.org/10.1016/j.jas.2008.04.023 less intensive occupation of the site at that time. The shift from the dominance of D. gigas to T. sarmaticus may have been caused by rising 8. Henshilwood CS, d’Errico F, Van Niekerk KL, Coquinot Y, Jacobs Z, Lauritzen sea levels, resulting in environmental changes from sand-covered rocky SE, et al. A 100,000-year-old ochre-processing workshop at Blombos shores prior to 11 ka and more exposed rocks thereafter. The absence of Cave, South Africa. Science. 2011;334:219–222. https://doi.org/10.1126/ C. meridionalis at KDC, which is present at both NBC9 and MRS19 during science.1211535 the Oakhurst, and the rarity of P. perna, may be due to unsuitable habitats 9. Klein RG. Preliminary report on the July through September 1970 excavations for these species at KDC, and not related to sea surface temperatures. at Nelson Bay Cave, Plettenberg Bay (Cape Province, South Africa). Palaeoecol Effective sea surface temperature indicator shellfish species are not Afr. 1972;6:177–208. present at KDC but the absence of cold temperature species suggests 10. Lombard M, Wadley L, Deacon J, Wurz S, Parsons I, Mohapi M, et al. that sea surface temperatures were relatively warm. The terrestrial South African and Lesotho Stone Age sequence updated (I). S Afr Archaeol climate during the Oakhurst at KDC was most likely warm and arid. Bull. 2012;67(195):120–144. The sizes of T. sarmaticus opercula and C. oculus from the Oakhurst levels 11. Loftus E, Sealy J, Lee-Thorp J. New radiocarbon dates and Bayesian models for at KDC and NBC are smaller than those from MSA sites. The opercula are Nelson Bay Cave and Byneskranskop 1: Implications for the South African later larger than in the post-Oakhurst LSA sites. A combination of factors may stone age sequence. Radiocarbon. 2016;58:365–381. https://doi.org/10.1017/ explain this scenario better than a single cause. RDC.2016.12

Research Article Volume 115| Number 9/10 https://doi.org/10.17159/sajs.2019/5578 7 September/October 2019 Oakhurst at Klipdrift Cave and shellfish exploitation Page 8 of 9

12. Jerardino A. Mega shell middens and hunter-gatherer resource intensification 33. Kilburn R, Rippey E. Sea shells of southern Africa. Johannesburg: Macmillan along the West Coast of South Africa. In: Bailey GN, Hardy K, Abdoulaye C, South Africa; 1982. editors. Shell energy: Mollusc shells as coast resources. Oxford: Oxbow Books; 2013. p. 207–228. 34. Jerardino A, Marean CW. Shellfish gathering, marine paleoecology and modern human behavior: Perspectives from Cave PP13B, Pinnacle Point, South Africa. 13. Ryano KP, Wurz S, Van Niekerk KL, Henshilwood CS. The technology of the J Hum Evol. 2010;59:412–424. https://doi.org/10.1016/j.jhevol.2010.07.003 early Oakhurst lithic techno-complex from Klipdrift Cave, southern Cape, South Africa. Afr Archaeol Rev. 2017;34(1):93–119. https://doi.org/10.1007/ 35. Branch G, Branch M. The living shores of southern Africa. Cape Town: s10437-017-9246-2 Struik; 1981.

14. Mitchell P. The archaeology of southern Africa. Cambridge: Cambridge 36. Jerardino A. Shell density as proxy for reconstructing prehistoric aquatic University Press; 2002. resource exploitation: Perspectives from southern Africa. J Archaeol Sci Rep. 2015;6:637–644. 15. Deacon J. The Later Stone Age of southernmost Africa. Cambridge Monographs in African Archaeology. Oxford: BAR Publishing; 1984. 37. McLachlan A, Lombard HW. Growth and production in exploited and unexploited populations of a rocky shore gastropod, Turbo sarmaticus. The Veliger. 16. Wadley L. The Early Holocene layers of Rose Cottage Cave, eastern 1981;23:221–229. Free State: Technology, spatial patterns and environment. S Afr Archaeol Bull. 2000;55:18–31. https://doi.org/10.2307/3888889 38. Langejans GH, Van Niekerk KL, Dusseldorp GL, Thackeray JF. Middle Stone Age shellfish exploitation: Potential indications for mass collecting and resource 17. Wurz S. Human evolution, archaeology and the South African landscape during intensification at Blombos Cave and Klasies River, South Africa. Quat Int. the last 100 000 years. In: Knight J, Rogerson CM, editors. The geography of 2012;270:80–94. https://doi.org/10.1016/j.quaint.2011.09.003 South Africa – Contemporary changes and new directions. World Regional Geography Book Series. Cham: Springer; 2019. p. 125–132. https://doi. 39. Clark BC, Griffiths CL. Ecological energetic of mussels Choromytilus org/10.1007/978-3-319-94974-1_13 meridionalis under simulated intertidal rock pool condition. J Exp Mar Biol Ecol. 1990;137:63–77. https://doi.org/10.1016/0022-0981(90)90060-P 18. Klein RG. The late Quaternary mammalian fauna of Nelson Bay Cave (Cape Province, South Africa): Its implications for megafaunal extinctions and 40. Marshall DJ, McQuaid CD. Differential physiological and behavioural responses environmental and cultural change. Quat Res. 1972;2(2):135–142. https://doi. of the intertidal mussels, Choromytilus meridionalis (Kr.) and Perna perna L., to org/10.1016/0033-5894(72)90034-8 exposure to hypoxia and air: A basis for spatial separation. J Exp Mar Biol Ecol. 1993;171:225–237. https://doi.org/10.1016/0022-0981(93)90005-9 19. Döckel W. Re-investigation of the Matjes River Rock Shelter. Stellenbosch: Stellenbosch University; 1998. 41. Bustamante RH, Branch GM. Large scale patterns and trophic structure of southern African rocky shores: The roles of geographic variation and wave 20. Schweitzer FR, Wilson ML. Byneskranskop 1: A late Quaternary living site in exposure. J Biogeogr. 1996;23(3):339–351. https://doi.org/10.1046/j.1365- the southern Cape Province, South Africa. Ann S Afr Mus. 1982;88:1–203. 2699.1996.00026.x

21. Goodwin AJH. Archaeology of the Oakhurst Shelter. Part 1: Course 42. Zardi GI, Nicastro KR, McQuaid CD, Erlandsson J. Sand and wave induced of the excavation. Trans R Soc S Afr. 1938;25:229–247. https://doi. mortality in indigenous (Perna perna) and invasive (Mytilus galloprovincialis) org/10.1080/00359193709519751 mussels. Mar Biol. 2008;153:853–858. https://doi.org/10.1007/s00227-007- 22. Deacon J. Wilton: An assessment after fifty years. S Afr Archaeol Bull. 0857-z 1972;10:10–48. https://doi.org/10.2307/3888813 43. Jerardino A, Branch GM, Navarro R. Human impact on precolonial West Coast 23. Deacon HJ. Where hunters gathered: A study of Holocene Stone Age people marine environments of South Africa. In: Erlandson JM, Rick TC, editors. in the eastern Cape. Cape Town: South African Archaeological Society; 1976. Human impacts on marine environments. Berkeley, CA: University of California Press; 2008. p. 279–296. 24. Schrire C. Oakhurst: Re-examination and vindication. S Afr Archaeol Bull. 1962;17:181–195. https://doi.org/10.2307/3887544 44. Foster GG, Hodgson AN, Balarin M. Effect of diet on growth rate and reproductive fitness of Turbo sarmaticus (: : ). Mar Biol. 25. Klein RG, Steele TE. Archaeological shellfish size and later human evolution 1999;134:307–315. https://doi.org/10.1007/s002270050548 in Africa. Proc Natl Acad Sci USA. 2013;110:10910–10915. https://doi. org/10.1073/pnas.1304750110 45. Ryano KP. The Later Stone Age in the southern Cape, South Africa, during the Terminal Pleistocene/Early Holocene with a focus on Klipdrift Cave [PhD 26. Klein RG, Bird DW. Shellfishing and human evolution. J Anthropol Archaeol. thesis]. Johannesburg: University of the Witwatersrand; 2014. 2016;44(Part B):198–205. 46. Henshilwood CS, Sealy JC, Yates R, Cruz-Uribe K, Goldberg P, Grine FE, et 27. Steele TE, Klein RG. Mollusk and tortoise size as proxies for stone age al. Klein Blombos Cave, southern Cape, South Africa: Preliminary report on population density in South Africa: Implications for the evolution of human the 1992-1999 excavations of the Middle Stone Age levels. J Archaeol Sci. cultural capacity. Munibe Antropologia-Arkeologia. 2005;57(1):221–237. 2001;28:421–448. https://doi.org/10.1006/jasc.2000.0638

28. Sealy J, Galimberti M. Shellfishing and the interpretation of shellfish sizes 47. Jerardino A. Large shell middens in Lamberts Bay, South Africa: A case of in the Middle and Later Stone Ages of South Africa. In: Bicho NF, Haws JA, hunter-gatherer resource intensification. J Archaeol Sci. 2010;37:2291– Davis L, editors. Trekking the shore: Changing coastlines and the antiquity 2302. https://doi.org/10.1016/j.jas.2010.04.002 of coastal settlements. New York: Springer; 2011. p. 405–419. https://doi. org/10.1007/978-1-4419-8219-3_17 48. Jerardino A. Prehistoric exploitation of marine resources in southern Africa with particular reference to shellfish gathering: Opportunities and continuities. 29. Giovas CM, Fitzpatrick SM, Stone J. Intensifying collection and size increase Pyrenae. 2010;1(41):7–52. tessellated nerite snail (Nerita tessellata) at the Coconut Walk site, Nevis, northern Lesser Antilles, AD 890–1440. J Archaeol Sci. 2013;40:4024–4038. 49. McQuaid CD, Branch GM. Influence of sea temperature, substratum and wave https://doi.org/10.1016/j.jas.2013.05.008 exposure on rock intertidal communities: An analysis of faunal and floral biomass. Mar Ecol. 1984;19:145–151. https://doi.org/10.3354/meps019145 30. Buchanan WF. Sea shells ashore: A study of the role of shellfish in prehistoric diet and lifestyle at Eland’s Bay, southwestern Cape, South Africa. Cape Town: 50. Sigman DM, Hain MP. The biological productivity of the ocean. Nature University of Cape Town; 1986. Education Knowledge. 2012;3(6):4.

31. Langejans GHJ, Dusseldorp GL, Van Niekerk K, Henshilwood CS. A hazy 51. Sachs JP, Anderson RF. Increased productivity in the subantarctic ocean during shade of winter: late Pleistocene environments and behavioural adaptations Heinrich events. Nature. 2005;434:1118. https://doi.org/10.1038/nature03544 at Blombos Cave, South Africa. In: Runge J, editor. New studies on former and 52. Foster GG, Hodgson AN. Intertidal population structure of the edible recent landscape changes in Africa. New York: CRC Press; 2013. p. 19–51. mollusc Turbo sarmaticus (Vetigastropoda) at an unexploited and exploited 32. Henshilwood CS. Holocene prehistory of the southern Cape, South Africa. sites along the coast of the Eastern Cape Province, South Africa. Afr Zool. Cambridge Monographs in African Archaeology. Oxford: BAR Publishing; 2008. 2000;35(2):173–183. https://doi.org/10.1080/15627020.2000.11657088

Research Article Volume 115| Number 9/10 https://doi.org/10.17159/sajs.2019/5578 8 September/October 2019 Oakhurst at Klipdrift Cave and shellfish exploitation Page 9 of 9

53. Proudfoot L, Kaehler S, McGarry DK, Uppink PA, Aereboe M, Morris KM. 57. Deacon J. Changing patterns in the Late Pleistocene/Early Holocene Exploitation status of infralittoral abalone (Haliotis midae) and alikreukel (Turbo prehistory of southern Africa as seen from the Nelson Bay Cave stone artifact sarmaticus) in the southern section of the Eastern Cape coast, South Africa. S sequence. Quat Res. 1978;10:84–111. https://doi.org/10.1016/0033-5894­ Afr J Sci. 2006;102:162–168. (78)­9­0­015-7

54. Halkett D, Hart T, Yates R, Volman TP, Parkington JE, Orton J, et al. First 58. Clark PU, Dyke AS, Shakun JD, Carlson AE, Clark J, Wohlfarth B, et al. The excavation of intact Middle Stone Age layers at Ysterfontein, Western Cape Last Glacial Maximum. Science. 2009;325:710–714. https://doi.org/10.1126/ Province, South Africa: Implications for Middle Stone Age ecology. J Archaeol science.1172873 Sci. 2003;30:955–971. https://doi.org/10.1016/S0305-4403(02)00273-X 55. Parkington J. Middens and moderns: Shellfishing and the Middle Stone Age 59. Thackeray JF, Scott L. The Younger Dryas in the Wonderkrater sequence, of the Western Cape. S Afr J Sci. 2003;99:243–247. South Africa? Ann Transv Mus. 2006;43:111–112. 56. Tonner TWW. Later Stone Age shellfishing behaviour at Dunefield Midden 60. Fitchett JM, Grab SW, Bamford MK, Mackay AW. Late Quaternary research (Western Cape, South Africa). J Archaeol Sci. 2005;32:1390–1407. https:// southern Africa: Progress, challenges and future trajectories. Trans R Soc S Afr. doi.org/10.1016/j.jas.2005.03.017 2017;72(3):280–293. https://doi.org/10.1080/0035919X.2017.1297966

Research Article Volume 115| Number 9/10 https://doi.org/10.17159/sajs.2019/5578 9 September/October 2019